M3NAS: Multi-Scale and Multi-Level Memory-Efficient Neural Architecture Search for Low-Dose CT Denoising.

Lowering the radiation dose in computed tomography (CT) can greatly reduce the potential risk to public health. However, the reconstructed images from dose-reduced CT or low-dose CT (LDCT) suffer from severe noise which compromises the subsequent diagnosis and analysis. Recently, convolutional neural networks have achieved promising results in removing noise from LDCT images. The network architectures that are used are either handcrafted or built on top of conventional networks such as ResNet and U-Net. Recent advances in neural network architecture search (NAS) have shown that the network architecture has a dramatic effect on the model performance. This indicates that current network architectures for LDCT may be suboptimal. Therefore, in this paper, we make the first attempt to apply NAS to LDCT and propose a multi-scale and multi-level memory-efficient NAS for LDCT denoising, termed M3NAS. On the one hand, the proposed M3NAS fuses features extracted by different scale cells to capture multi-scale image structural details. On the other hand, the proposed M3NAS can search a hybrid cell- and network-level structure for better performance. In addition, M3NAS can effectively reduce the number of model parameters and increase the speed of inference. Extensive experimental results on two different datasets demonstrate that the proposed M3NAS can achieve better performance and fewer parameters than several state-of-the-art methods. In addition, we also validate the effectiveness of the multi-scale and multi-level architecture for LDCT denoising, and present further analysis for different configurations of super-net.

DGRUnit: Dual graph reasoning unit for brain tumor segmentation.

Many fully automatic segmentation models have been created to solve the difficulty of brain tumor segmentation, thanks to the rapid growth of deep learning. However, few approaches focus on the long-range relationships and contextual interdependence in multimodal Magnetic Resonance (MR) images. In this paper, we propose a novel approach for brain tumor segmentation called the dual graph reasoning unit (DGRUnit). Two parallel graph reasoning modules are included in our proposed method: a spatial reasoning module and a channel reasoning module. The spatial reasoning module models the long-range spatial dependencies between distinct regions in an image using a graph convolutional network (GCN). The channel reasoning module uses a graph attention network (GAT) to model the rich contextual interdependencies between different channels with similar semantic representations. Our experimental results clearly demonstrate the superior performance of the proposed DGRUnit. The ablation study shows the flexibility and generalizability of our model, which can be easily integrated into a wide range of neural networks and further improve them. When compared to several state-of-the-art methods, experimental results show that the proposed approach significantly improves both visual inspection and quantitative metrics for brain tumor segmentation tasks.

Training low dose CT denoising network without high quality reference data.

Objective.Currently, the field of low-dose CT (LDCT) denoising is dominated by supervised learning based methods, which need perfectly registered pairs of LDCT and its corresponding clean reference image (normal-dose CT). However, training without clean labels is more practically feasible and significant, since it is clinically impossible to acquire a large amount of these paired samples. In this paper, a self-supervised denoising method is proposed for LDCT imaging.Approach.The proposed method does not require any clean images. In addition, the perceptual loss is used to achieve data consistency in feature domain during the denoising process. Attention blocks used in decoding phase can help further improve the image quality.Main results.In the experiments, we validate the effectiveness of our proposed self-supervised framework and compare our method with several state-of-the-art supervised and unsupervised methods. The results show that our proposed model achieves competitive performance in both qualitative and quantitative aspects to other methods.Significance.Our framework can be directly applied to most denoising scenarios without collecting pairs of training data, which is more flexible for real clinical scenario.

Extreme Low-Resolution Activity Recognition Using a Super-Resolution-Oriented Generative Adversarial Network.

Activity recognition is a fundamental and crucial task in computer vision. Impressive results have been achieved for activity recognition in high-resolution videos, but for extreme low-resolution videos, which capture the action information at a distance and are vital for preserving privacy, the performance of activity recognition algorithms is far from satisfactory. The reason is that extreme low-resolution (e.g., 12 × 16 pixels) images lack adequate scene and appearance information, which is needed for efficient recognition. To address this problem, we propose a super-resolution-driven generative adversarial network for activity recognition. To fully take advantage of the latent information in low-resolution images, a powerful network module is employed to super-resolve the extremely low-resolution images with a large scale factor. Then, a general activity recognition network is applied to analyze the super-resolved video clips. Extensive experiments on two public benchmarks were conducted to evaluate the effectiveness of our proposed method. The results demonstrate that our method outperforms several state-of-the-art low-resolution activity recognition approaches.